RGB encoder probs

[Arakon]

I got a selfbuilt RGB encoder here, based off the AD 724. the problem is that I have horizontal, thick lines running through the pic, mostly noticable on black background, totally invisible on white. the s-video picture is VERY good despite the lines, the AV pic good, but gets distorted by the lines.

the encoder is built exactly by the specs (example circuit). the s-video and av cables give clear pic with other devices. the above happens with both the power off the jamma harness (5.1 V) and using a seperate power supply giving out exactly 5V.
the cables are kept short between the pcb and the ports as well.

any ideas?

<small>[ August 14, 2003, 08:24 PM: Message edited by: Arakon ]</small>

 

[JAMMASLOT]

Humm, just try to help you..

- try with the last AD725
- you need power line filter (50/60Hz osc?)
- add a lm1881 befor the AD724 (synch prob?)



<small>[ August 14, 2003, 08:26 PM: Message edited by: JAMMASLOT ]</small>

 

[Arakon]

I can't get a 725 right now.
the PSU does provide the filtering, as far as I know.
the sync is clear mvs composite sync, there wouldn't be a sense in using a lm1881 I believe.

 

[JAMMASLOT]

ok, just try to help u, i see you're really happy of this blow_top

I mean Filtering the Mains line ...
glee

If you're in germany you can get ad725 here, and ship to your country, using the same money.
spock

 

[Reznor007]

Might be a poor ground on your encoder somewhere. I built an encoder using a 725 and I could only get it to work reliably with my SF2 CE PCB. With my MK2 PCB the color would be beyond messed up. I just went with the JROK after a while.

If you still want to try the home built, you can get 2 free 725 from Analog Devices. Just say you are from a company and they will send you 2 free samples(I got mine within 3 days of emailing).

 

[Arakon]

unfortunately I already got the 724 a year ago as sample, also, they only ship to US addresses.
the grounds are all ok.

as for you having probs with MK2, not surprising.. it's synch is out of spec, it causes sony TVs to go crazy too.

 

[Reznor007]

You might try different value capacitors and resistors. I had to tweak my setup a bit as the color was overdriven on SF2(some things would go white because they were too bright). Are your digital and analog grounds separate?

As for MK2, it was only the color, the sync was perfectly on. I think the problem had to do with my subchannel carrier section. It works great with the JROK.

 

[Arakon]

the 724 datasheet only refers to a single ground connection for all, and all signals are analog anyways (RGB in, s-video out).
I did play with the RGB input level some (using pots), other than adjusting the color brightness it had no effect.

 

[JAMMASLOT]

continue .... you'll find the answer

 

[Reznor007]

Arakon:
the 724 datasheet only refers to a single ground connection for all, and all signals are analog anyways (RGB in, s-video out).
I did play with the RGB input level some (using pots), other than adjusting the color brightness it had no effect.

By analog and digital ground I mean ground for the RGB channels and Y/C output, and ground for the chip logic(considered digital ground).

 

[JAMMASLOT]

Reznor007:

Arakon:
the 724 datasheet only refers to a single ground connection for all, and all signals are analog anyways (RGB in, s-video out).
I did play with the RGB input level some (using pots), other than adjusting the color brightness it had no effect.

By analog and digital ground I mean ground for the RGB channels and Y/C output, and ground for the chip logic(considered digital ground).

I think it's also what analog devices talks about...

 

[Matt]

Try using a new power supply. A junky PSU can give exactly the effect you're describing.

Adding a voltage regulator on the circuits +5v input might help too.

 

[Arakon]

alright, then what do I do with analog positive supply and digital positive supply? digital would be the normal 5V line off the jamma harness.. but analog? I can split the ground connections (jamma GND and video GND), although I think they are connected anyways.

 

[Reznor007]

Definitely separate the grounds. Use the specified video ground as well. For the analog/digital power inputs, run a separate line from the PSU for each one. For pins that need to be logic high(such as chip enable) just split that off the digital power in.

 

[Arakon]

@matt: I did, in fact several.. jamenco PSU, selfbuilt PSU that outputs a clean 5.0 V, and the 5V regulator on the 2-slot.

 

[Arakon]

@reznor:

I did, but it made no diff at all.. also, on my 2-slots the grounds are 100% connected (video gnd and main gnd).

 

[hardware_hack]

In terms of seperate grounds, those most likely aren't important in your case since you built the example circuit they gave. It's only really critical when you have other transient loads on the rails. Just keep every lead as short as possible, and make sure you use bypass caps!!! Much more important than seperate grounds. There might be oscillation at some harmonic of the ntsc frequency screwing up the dc restore circuit, which is aliasing on the display so it looks like it is creeping slowly.

Have you tried different clocking schemes? Using a 4fc crystal, etc. Looking at the waveforms on a scope?

How fast do the lines move/how thick are they? It sounds like a problem I was having with the black level clamp when I built mine out of discretes, but the problem is that part is totally inaccessable in the 724.

Just throwing ideas out. Good luck.

 

[Arakon]

got no scope unfortunately, and didn't try another crystal yet (none handy).. the odd part is that the lines seem to not move, they remain in the same place, and then have lines "flowing" from top to bottom of each line.

the speed of that seems to change with the colors displayed, and yes, it has a "creeping" effect.
the lines are half an inch thick, maybe a bit more. they're mostly visible on black, but also on red.. totally gone on white.

 

[hardware_hack]

Why not get some color bars up and see which colors are affected?

 

[Arakon]

the colors themselves are flawless, the picture except for the interference is perfectly sharp too.
the brighter the picture gets, the least the interference is visible.. so basically, it's "dark"/black that shows them the most.

<small>[ August 15, 2003, 01:54 PM: Message edited by: Arakon ]</small>

 

[NEOGEOman]

It sounds like a power or ground problem. You're dealing with some pretty intolerant things here, and if you've got sloppy soldering or too-thin wires or too much distance between certain components you can get what you describe. It's not an easy thing, building your own encoder.

 

[Reznor007]

NEOGEOman:
It sounds like a power or ground problem. You're dealing with some pretty intolerant things here, and if you've got sloppy soldering or too-thin wires or too much distance between certain components you can get what you describe. It's not an easy thing, building your own encoder.

So true. I originally saw the JROK for $75+shipping, so I thought, hell, I can do it for 1/4 that price. By the time I had my encoder working flawlessly with SF2, I had spent nearly $40 on connectors, components, a bare PCB, and so on. When you think about the time and effort involved in building it, the JROK sounds better all the time.

 

[Arakon]

the thing is, I don't NEED it. I'd still like to get it running. but not for 90-100 bucks (to germany).
I had the components around, no extra cost there. I'm probably receiving a 725 now, but that can take a month according to AD :/

 

[Reznor007]

Make sure you have a crystal fast enough for the 725, it requires 4x subchannel carrier clock. 14.?? for NTSC and 17.?? for PAL (Can't remember exact values).

I had to order my crystal from Jameco as Radio didn't have it, and when I tried to have them order one in, they had discontinued it(and almost all other electronic components...Radio Shack is becoming a Best Buy wanna-be instead of a hobby store).

 

[JAMMASLOT]

NEOGEOman:
It sounds like a power or ground problem. You're dealing with some pretty intolerant things here, and if you've got sloppy soldering or too-thin wires or too much distance between certain components you can get what you describe. It's not an easy thing, building your own encoder.

true, video signal is High frequency signal, be carefull of wires, compos...